import os import six import yaml import copy import numpy as np from lib.utils.collections import AttrDict from lib.utils.misc import get_run_name from ast import literal_eval __C = AttrDict() # Consumers can get config by: cfg = __C # Root directory of project __C.ROOT_DIR = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) # ---------------------------------------------------------------------------- # # Data configurations # ---------------------------------------------------------------------------- # __C.DATASET = AttrDict() __C.DATASET.NAME = 'nyu' __C.DATASET.RGB_PIXEL_MEANS = (0.485, 0.456, 0.406) # (102.9801, 115.9465, 122.7717) __C.DATASET.RGB_PIXEL_VARS = (0.229, 0.224, 0.225) # (1, 1, 1) # Scale the depth map __C.DATASET.DEPTH_SCALE = 10.0 __C.DATASET.CROP_SIZE = (385, 385) # (height, width) # Minimum depth after data augmentation __C.DATASET.DEPTH_MIN = 0.001 # Maximum depth
__C.BBOX_XFORM_CLIP = np.log(1000. / 16.) # Pixel mean values (BGR order) as a (1, 1, 3) array # We use the same pixel mean for all networks even though it's not exactly what # they were trained with # "Fun" fact: the history of where these values comes from is lost (From Detectron lol) __C.PIXEL_MEANS = np.array([[[102.9801, 115.9465, 122.7717]]]) # For reproducibility __C.RNG_SEED = 3 # A small number that's used many times __C.EPS = 1e-14 # Root directory of project __C.ROOT_DIR = osp.abspath(osp.join(osp.dirname(__file__), '..', '..')) # Output basedir __C.OUTPUT_DIR = 'Outputs' # Name (or path to) the matlab executable __C.MATLAB = 'matlab' # Dump detection visualizations __C.VIS = False # Score threshold for visualization __C.VIS_TH = 0.9 # Expected results should take the form of a list of expectations, each # specified by four elements (dataset, task, metric, expected value). For